A recurring issue in psychotropic prescribing is the difficulty in predicting which patients will respond to a given drug, and which will experience side effects. Physicians could respond to such heterogeneity by customizing their prescribing and using different medications for different patients. However, other factors such as formularies may work against customization. It is not well known how much customization is occurring, how its use varies among physicians, and what characteristics of physicians and patients explain the variation. For example, specialist physicians such as psychiatrists may be more likely to customize than general practitioners, and patients with more comorbidities may require more customization. This study aims to improve our understanding of customization in medication selection for two disorders: major depression and bipolar disorder. The specific aims are to: 1. Examine how the extent of customization in prescribing is associated with key aspects of pharmacotherapy adequacy, such as the proportion of patients achieving recommended duration of appropriate medications, and with patient outcomes. 2. Examine how the extent of customization in prescribing is associated with total expenditures for psychotropic medications, and for other health care. 3. Measure how the extent of customization is affected by characteristics of physicians and patients, and by system interventions such as changes in formulary composition. To address these aims, the applicants propose to study psychotropic prescribing patterns in three settings: MHN (a national managed behavioral health organization);the dually eligible Medicaid-Medicare populations in Michigan (MI) and Indiana (IN);and a naturalistic clinical effectiveness trial (STEP-BD). The results of this study will have important implications for health plans and other decision makers in behavioral health care. Understanding the extent of customization in psychotropic prescribing, its determinants, and its relationship with adequacy of treatment will facilitate strategies for positively influencing physician behavior and improving patient care and outcomes. If research identifies patient or physician predictors of poor customization, policy can target educational efforts or incentives based on those predictors. Poor adherence to psychotropic medications is a pervasive problem, which could potentially be reduced by better matching of medications to patients (customization). This research seeks to measure the extent of customization and see how it relates to treatment adequacy, outcomes and costs. If we identify patient or physician predictors of poor customization, these could be used by health plans and public payers to develop incentives or educational programs to improve treatment quality by promoting more customization in prescribing.